Embody4D: A Generalist 4D World Model for Embodied AI

TL;DR

Embody4D introduces a novel 4D world model for embodied AI that synthesizes consistent multi-view videos from monocular inputs, overcoming data scarcity and fidelity challenges to enhance robotic planning.

Contribution

The paper presents Embody4D, a comprehensive 4D world model with a new data synthesis pipeline, adaptive regularization, and interaction-aware attention, advancing embodied spatial reasoning.

Findings

Achieves state-of-the-art view synthesis quality.

Demonstrates improved robotic planning and learning.

Ensures spatiotemporal consistency in generated videos.

Abstract

World models have made significant progress in modeling dynamic environments; however, most embodied world models are still restricted to 2D representations, lacking the comprehensive multi-view information essential for embodied spatial reasoning. Bridging this gap is non-trivial, primarily due to challenges from severe scarcity of paired multi-view data, the difficulty of maintaining spatiotemporal consistency in generated 3D geometries, and the tendency to hallucinate manipulation details. To address these challenges, we propose Embody4D, a dedicated video-to-video world model for embodied scenarios, capable of synthesizing arbitrary novel views from a monocular video. First, to tackle data scarcity, we introduce a 3D-aware compositional synthesis pipeline to curate a heterogeneous dataset compositing cross-embodiment robotic arms with diverse backgrounds, guaranteeing broad generalization. Second, to enforce geometric stability, we devise an adaptive noise injection strategy; by leveraging confidence disparities across image regions, this method selectively regularizes the diffusion process to ensure strict spatiotemporal consistency. Finally, to guarantee manipulation fidelity, we incorporate an interaction-aware attention mechanism that explicitly attends to the robotic interaction regions. Extensive experiments demonstrate that Embody4D achieves state-of-the-art performance, serving as a robust world model that synthesizes high-fidelity, view-consistent videos to empower downstream robotic planning and learning.